UCLA

Compared to other disciplines, one of the distinctive features of Economics is the impossibility of doing experiments to study empirically relevant questions. In this dissertation, I use two episodes from the past where the unique features of the institutional environment created quasi-experiments that help us understand relevant issues relate to cartels and neighborhood effects. In the fist two chapters, I use the Chilean nitrate industry between the War of the Pacific and the start of the First World War (1880-1914) to shed light on the effects of entry for cartels and the importance of learning for effective cartel organization. The Chilean nitrate industry was very important at the time: Chilean nitrate was the main commercial fertilizer used in the world. Also, it was the main industry of Chile, the only country where it is found, where it represented 70% of exports and 45% of government revenues. Importantly, there was no antitrust legislation and no domestic consumer surplus to protect in Chile. Thus, cartels could be freely formed by nitrate producers. Moreover, these cartels were completely public: Their decisions were publicly discussed by the press and the public and they would be formed by the signature of a public contract. I collected from handwritten archival records an original data set of monthly output and inputs that covers 35 years of this industry, period over which 5 cartels were organized. In the third chapter, I introduce a novel new dataset to study neighborhood effects that uses a massive program of forced-displacement of population within the city of Santiago in the context of the Chilean dictatorship in the early 1980s.

Chapter 1 studies the effect of cartels on entry, and the long-term effect of cartel-induced entry for the evolution of industry productivity. Intuitively, as cartels artificially increase prices entry becomes more attractive and less productive firms may decide to enter. Because of data limitations, previous researchers have not quantified the extensive margin mechanism. My analysis has 3 steps. First, I evaluate if nitrate cartels caused more entry of new firms. I find that cartels generated an additional entry of 4 plants per year or about 5\% of the initial number of plants in my main period of interest. Second, I estimate the productivity of all the plants in the industry to analyze if firms that entered during cartel periods where less productive. I find that there is a sizable gap in average productivity between entrants during competition and cartel periods: If the median-sized plant in the industry had received this productivity gap its revenue would have increases by one-third. Third, I conduct two counterfactual simulations. The first, studies the effect of entry on cartel profits by estimating incumbent cartel members counterfactual profits if they had been able to prevent any additional entry. I find their counterfactual profits would have been 40% larger had there been no entry. The second counterfactual studies the effect of having a cartel on entry. Specifically, what would the number of new firms and their productivity have been, if during the Fourth and Fifth Cartels there had been competition? I find that 18% of the plants that entered in the data would have not entered. This translates into an increase on mean plant productivity of 3%.

Chapter 2 studies the degree to which experience helps firms to organize successful cartels. Unlike in most of the previous literature on cartels, in the nitrate cartels issues related to monitoring and enforcement were of secondary importance with respect to the challenge of allocating the collusive surplus among the colluding firms. I document that cartel contracts gradually became more complete, generated a smoother transition from competition to collusion, and that producers eventually discarded inefficient methods of market share allocation, associated to larger production costs, in favor of better alternatives. In particular, I am able to estimate that the time method used during the Second Cartel directly caused higher production costs of about 10%, while the trial method implemented during the Third Cartel caused higher costs of approximately 20%.

Finally, Chapter 3 describes a novel large dataset that combines archival records and administrative data to study a natural experiment that occurred during the Chilean dictatorship between 1979 and 1985, when the government mandated the relocation of a large number of slums in the city of Santiago, Chile. Some features of the program's implementation make it of unique interest to study the broad effects of neighborhoods on social mobility and inequality: the unit of treatment was the slum, participation was mandatory and compliance was very high, since the policy was implemented during a highly repressive dictatorial government. In addition, and only some of the slums were removed from their original location creating two groups of families: movers and non-movers, which allows me to identify a causal displacement effect. The dataset comprises data for more than 26,000 households that were part of this program (out of a total of 40,000 households) and more than 58,000 of their children, providing the potential for causal estimation of the long-term and inter-generational effects of moving to a high-poverty neighborhood on education, mortality, income, and crime.